lib/ignore: Revert comma handling, upgrade globbing package

This was fixed upstream due to our ticket, so we no longer need the
manual handling of commas. Keep the tests and better debug output around
though.

GitHub-Pull-Request: https://github.com/syncthing/syncthing/pull/3081

lib/ignore/ignore.go

@@ -337,7 +337,7 @@ func parseIgnoreFile(fd io.Reader, currentFile string, seen map[string]bool) ([]
 			continue
 		}
 
-		line = escapeCommas(filepath.ToSlash(line))
+		line = filepath.ToSlash(line)
 		switch {
 		case strings.HasPrefix(line, "#"):
 			err = addPattern(line)
@@ -358,49 +358,3 @@ func parseIgnoreFile(fd io.Reader, currentFile string, seen map[string]bool) ([]
 
 	return patterns, nil
 }
-
-// escapes unescaped commas encountered outside of brackets
-func escapeCommas(s string) string {
-	buf := make([]rune, 0, len(s))
-	inEscape := false
-	inBrackets := 0
-	inSquareBrackets := 0
-	for _, r := range s {
-		// Escaped characters are passed on verbatim no matter what, and we
-		// clear the escape flag for the next character.
-		if inEscape {
-			buf = append(buf, r)
-			inEscape = false
-			continue
-		}
-
-		// Check for escapes and commas to escape. Also keep track of the
-		// brackets level by counting start and end brackets of the two
-		// types.
-
-		switch r {
-		case '\\':
-			inEscape = true
-
-		case '{':
-			inBrackets++
-		case '}':
-			inBrackets--
-		case '[':
-			inSquareBrackets++
-		case ']':
-			inSquareBrackets--
-
-		case ',':
-			// Commas should be escaped if we're not inside a brackets
-			// construction, and if they weren't already escaped (in which
-			// case we'll have taken the first branch way up top).
-			if inBrackets == 0 && inSquareBrackets == 0 {
-				buf = append(buf, '\\')
-			}
-		}
-
-		buf = append(buf, r)
-	}
-	return string(buf)
-}

vendor/github.com/gobwas/glob/lexer.go

@@ -9,7 +9,7 @@ import (
 
 const (
 	char_any           = '*'
-	char_separator     = ','
+	char_comma         = ','
 	char_single        = '?'
 	char_escape        = '\\'
 	char_range_open    = '['
@@ -138,7 +138,7 @@ type lexer struct {
 func newLexer(source string) *lexer {
 	l := &lexer{
 		input:       source,
-		state:       lexText,
+		state:       lexRaw,
 		items:       make(chan item, len(source)),
 		termPhrases: make(map[int]int),
 	}
@@ -146,7 +146,7 @@ func newLexer(source string) *lexer {
 }
 
 func (l *lexer) run() {
-	for state := lexText; state != nil; {
+	for state := lexRaw; state != nil; {
 		state = state(l)
 	}
 	close(l.items)
@@ -218,29 +218,26 @@ func (l *lexer) acceptAll(valid string) {
 	l.unread()
 }
 
-func (l *lexer) emit(t itemType) {
-	if l.pos == len(l.input) {
-		l.items <- item{t, l.input[l.start:]}
-	} else {
-		l.items <- item{t, l.input[l.start:l.pos]}
-	}
+func (l *lexer) emitCurrent(t itemType) {
+	l.emit(t, l.input[l.start:l.pos])
+}
 
+func (l *lexer) emit(t itemType, s string) {
+	l.items <- item{t, s}
 	l.start = l.pos
 	l.runes = 0
 	l.width = 0
 }
 
-func (l *lexer) emitMaybe(t itemType) {
-	if l.pos > l.start {
-		l.emit(t)
-	}
-}
-
 func (l *lexer) errorf(format string, args ...interface{}) {
 	l.items <- item{item_error, fmt.Sprintf(format, args...)}
 }
 
-func lexText(l *lexer) stateFn {
+func (l *lexer) inTerms() bool {
+	return len(l.termScopes) > 0
+}
+
+func lexRaw(l *lexer) stateFn {
 	for {
 		c := l.read()
 		if c == eof {
@@ -248,21 +245,8 @@ func lexText(l *lexer) stateFn {
 		}
 
 		switch c {
-		case char_escape:
-			l.unread()
-			l.emitMaybe(item_text)
-
-			l.read()
-			l.ignore()
-
-			if l.read() == eof {
-				l.errorf("unclosed '%s' character", string(char_escape))
-				return nil
-			}
-
 		case char_single:
 			l.unread()
-			l.emitMaybe(item_text)
 			return lexSingle
 
 		case char_any:
@@ -274,33 +258,35 @@ func lexText(l *lexer) stateFn {
 			}
 
 			l.unread()
-			l.emitMaybe(item_text)
 			return n
 
 		case char_range_open:
 			l.unread()
-			l.emitMaybe(item_text)
 			return lexRangeOpen
 
 		case char_terms_open:
 			l.unread()
-			l.emitMaybe(item_text)
 			return lexTermsOpen
 
 		case char_terms_close:
 			l.unread()
-			l.emitMaybe(item_text)
 			return lexTermsClose
 
-		case char_separator:
+		case char_comma:
+			if l.inTerms() { // if we are not in terms
+				l.unread()
+				return lexSeparator
+			}
+			fallthrough
+
+		default:
 			l.unread()
-			l.emitMaybe(item_text)
-			return lexSeparator
+			return lexText
 		}
 	}
 
 	if l.pos > l.start {
-		l.emit(item_text)
+		l.emitCurrent(item_text)
 	}
 
 	if len(l.termScopes) != 0 {
@@ -308,11 +294,41 @@ func lexText(l *lexer) stateFn {
 		return nil
 	}
 
-	l.emit(item_eof)
+	l.emitCurrent(item_eof)
 
 	return nil
 }
 
+func lexText(l *lexer) stateFn {
+	var escaped bool
+	var data []rune
+
+scan:
+	for c := l.read(); c != eof; c = l.read() {
+		switch {
+		case c == char_escape:
+			escaped = true
+			continue
+
+		case !escaped && c == char_comma && l.inTerms():
+			l.unread()
+			break scan
+
+		case !escaped && utf8.RuneLen(c) == 1 && special(byte(c)):
+			l.unread()
+			break scan
+
+		default:
+			data = append(data, c)
+		}
+
+		escaped = false
+	}
+
+	l.emit(item_text, string(data))
+	return lexRaw
+}
+
 func lexInsideRange(l *lexer) stateFn {
 	for {
 		c := l.read()
@@ -325,7 +341,7 @@ func lexInsideRange(l *lexer) stateFn {
 		case char_range_not:
 			// only first char makes sense
 			if l.pos-l.width == l.start {
-				l.emit(item_not)
+				l.emitCurrent(item_not)
 			}
 
 		case char_range_between:
@@ -338,8 +354,13 @@ func lexInsideRange(l *lexer) stateFn {
 			return lexRangeHiLo
 
 		case char_range_close:
+			if l.runes == 1 {
+				l.errorf("range should contain at least single char")
+				return nil
+			}
+
 			l.unread()
-			l.emitMaybe(item_text)
+			l.emitCurrent(item_text)
 			return lexRangeClose
 		}
 	}
@@ -362,7 +383,7 @@ func lexRangeHiLo(l *lexer) stateFn {
 				return nil
 			}
 
-			l.emit(item_range_between)
+			l.emitCurrent(item_range_between)
 
 		case char_range_close:
 			l.unread()
@@ -372,7 +393,7 @@ func lexRangeHiLo(l *lexer) stateFn {
 				return nil
 			}
 
-			l.emit(item_range_hi)
+			l.emitCurrent(item_range_hi)
 			return lexRangeClose
 
 		default:
@@ -385,35 +406,30 @@ func lexRangeHiLo(l *lexer) stateFn {
 				return nil
 			}
 
-			l.emit(item_range_lo)
+			l.emitCurrent(item_range_lo)
 		}
 	}
 }
 
 func lexAny(l *lexer) stateFn {
 	l.pos += 1
-	l.emit(item_any)
-	return lexText
+	l.emitCurrent(item_any)
+	return lexRaw
 }
 
 func lexSuper(l *lexer) stateFn {
 	l.pos += 2
-	l.emit(item_super)
-	return lexText
+	l.emitCurrent(item_super)
+	return lexRaw
 }
 
 func lexSingle(l *lexer) stateFn {
 	l.pos += 1
-	l.emit(item_single)
-	return lexText
+	l.emitCurrent(item_single)
+	return lexRaw
 }
 
 func lexSeparator(l *lexer) stateFn {
-	if len(l.termScopes) == 0 {
-		l.errorf("syntax error: separator not inside terms list")
-		return nil
-	}
-
 	posOpen := l.termScopes[len(l.termScopes)-1]
 
 	if l.pos-posOpen == 1 {
@@ -423,16 +439,16 @@ func lexSeparator(l *lexer) stateFn {
 
 	l.termPhrases[posOpen] += 1
 	l.pos += 1
-	l.emit(item_separator)
-	return lexText
+	l.emitCurrent(item_separator)
+	return lexRaw
 }
 
 func lexTermsOpen(l *lexer) stateFn {
 	l.termScopes = append(l.termScopes, l.pos)
 	l.pos += 1
-	l.emit(item_terms_open)
+	l.emitCurrent(item_terms_open)
 
-	return lexText
+	return lexRaw
 }
 
 func lexTermsClose(l *lexer) stateFn {
@@ -460,19 +476,19 @@ func lexTermsClose(l *lexer) stateFn {
 	delete(l.termPhrases, posOpen)
 
 	l.pos += 1
-	l.emit(item_terms_close)
+	l.emitCurrent(item_terms_close)
 
-	return lexText
+	return lexRaw
 }
 
 func lexRangeOpen(l *lexer) stateFn {
 	l.pos += 1
-	l.emit(item_range_open)
+	l.emitCurrent(item_range_open)
 	return lexInsideRange
 }
 
 func lexRangeClose(l *lexer) stateFn {
 	l.pos += 1
-	l.emit(item_range_close)
-	return lexText
+	l.emitCurrent(item_range_close)
+	return lexRaw
 }

vendor/github.com/gobwas/glob/lexer_test.go

@@ -16,6 +16,27 @@ func TestLexGood(t *testing.T) {
 				item{item_eof, ""},
 			},
 		},
+		{
+			pattern: "hello,world",
+			items: []item{
+				item{item_text, "hello,world"},
+				item{item_eof, ""},
+			},
+		},
+		{
+			pattern: "hello\\,world",
+			items: []item{
+				item{item_text, "hello,world"},
+				item{item_eof, ""},
+			},
+		},
+		{
+			pattern: "hello\\{world",
+			items: []item{
+				item{item_text, "hello{world"},
+				item{item_eof, ""},
+			},
+		},
 		{
 			pattern: "hello?",
 			items: []item{
@@ -124,12 +145,10 @@ func TestLexGood(t *testing.T) {
 		for i, exp := range test.items {
 			act := lexer.nextItem()
 			if act.t != exp.t {
-				t.Errorf("#%d wrong %d-th item type: exp: %v; act: %v (%s vs %s)", id, i, exp.t, act.t, exp, act)
-				break
+				t.Errorf("#%d %q: wrong %d-th item type: exp: %q; act: %q\n\t(%s vs %s)", id, test.pattern, i, exp.t, act.t, exp, act)
 			}
 			if act.s != exp.s {
-				t.Errorf("#%d wrong %d-th item contents: exp: %q; act: %q (%s vs %s)", id, i, exp.s, act.s, exp, act)
-				break
+				t.Errorf("#%d %q: wrong %d-th item contents: exp: %q; act: %q\n\t(%s vs %s)", id, test.pattern, i, exp.s, act.s, exp, act)
 			}
 		}
 	}

vendor/manifest

@@ -40,7 +40,7 @@
 		{
 			"importpath": "github.com/gobwas/glob",
 			"repository": "https://github.com/gobwas/glob",
-			"revision": "d877f6352135181470c40c73ebb81aefa22115fa",
+			"revision": "82e8d7da03805cde651f981f9702a4b4d8cf58eb",
 			"branch": "master"
 		},
 		{